Evidence of spillover of illness among household members: EQ-5D scores from a US sample

Abstract

Background/objectives: The effects of illness extend beyond the individual to caregivers and family members. This study identified evidence of spillover of illness onto household members' health-related quality of life.

Methods: Medical Expenditures Panel Survey (MEPS) data from 2000-2003 were analyzed using multivariable regression to identify spillover of household members' chronic conditions onto individuals' health-related quality of life as measured by the EuroQol-5D (EQ-5D) score (N = 24,188). Spillover was assessed by disease category, timing of occurrence (preexisting or new conditions), and age of the household member (adult or child).

Results: Controlling for an individual's own health conditions and other known predictors of EQ-5D scores, the authors found that the odds of an individual reporting full health (an EQ-5D score of 1.0, relative to <1.0) were lower with the presence of existing mental (odds ratio 0.71; 95% confidence interval, 0.64-0.79), respiratory (0.85; 0.75-0.97), and musculoskeletal (0.83; 0.75-0.93) conditions among adults and with mental (0.72; 0.62-0.82) and respiratory (0.80; 0.81-0.96) conditions among children in the household. The odds of an individual reporting full health were also lower for newly occurring chronic conditions in the household, including adults' mental (0.79; 0.65-0.97), nervous/sensory system (0.76; 0.61-0.96), and musculoskeletal (0.78; 0.65-0.95) conditions and children's mental conditions (0.64; 0.48-0.86). EQ-5D dimensions may be unsuited to fully capture spillover utility among household members, and MEPS lacks condition severity and caregiver status among household members.

Conclusions: Evidence from a US sample suggests that individuals who live with chronically ill household members have lower EQ-5D scores than those who live either alone or with healthy household members. Averting spillover effects may confer substantial additional benefit at the population level for interventions that prevent or alleviate conditions that incur such effects.

Figure 1

Static regression model results: stage 1 (logistic) and stage 2 (linear) models predicting association between respondent’s EQ-5D score and existing and new chronic conditions in other household members, by ICD9 category. Models are adjusted for respondents’ own chronic conditions and sociodemographic characteristics, and household characteristics (stage 1: n=24,188; stage 2: n=12,852).